Weyl differencing: as a difference sum
ProvedCircleMethod.weyl_squaringThe opening move of Weyl differencing. For any real-valued function on the natural numbers and any ,
The left-hand side is for the exponential sum; the right-hand side replaces the phase by its differences. When is a polynomial of degree , substituting makes the inner phase a polynomial of degree in for each fixed . Iterating the manoeuvre times reduces a degree- Weyl sum to linear sums, which the geometric bound already controls. Each squaring costs a square root in the final estimate, which is the origin of the exponent in Weyl's inequality.
Formalization Note. Stated as an equation in between and the double sum, rather than in using , so that it can be chained with other complex identities without casting. No hypotheses on or ; the case gives .
import Definitions.Def_CircleMethod_genfun import Mathlib.Analysis.SpecialFunctions.Pow.Real import Mathlib.Analysis.SpecialFunctions.Integrals.Basic import Mathlib.NumberTheory.Divisors import Mathlib.Data.Fintype.Pi open Finset
namespace CircleMethod
theorem weyl_squaring (φ : ℕ → ℝ) (N : ℕ) :
(∑ x ∈ Finset.range N, e (φ x)) * (starRingEnd ℂ) (∑ x ∈ Finset.range N, e (φ x))
= ∑ x ∈ Finset.range N, ∑ y ∈ Finset.range N, e (φ x - φ y) := by sorry
end CircleMethodRead-back
What the Lean code literally says, in plain math · claude-opus-5
The declaration weyl_squaring asserts the following, for every function assigning a real number to each natural number , and for every natural number (both universally quantified, with no hypotheses whatsoever imposed on either — is an arbitrary real-valued sequence, not assumed continuous, bounded, polynomial, or of any particular shape, and is arbitrary, including ):
Here the symbol is not a primitive: it is the additive character defined in this bundle by
a complex number of modulus , where the real argument is regarded as a complex number and is the real circle constant. The bar denotes complex conjugation on (the canonical conjugation ring automorphism of the complex numbers), applied to the complete sum; thus the left-hand side is the product of the sum with its own conjugate, i.e. the squared modulus of that sum, written as a product rather than as .
Both index ranges are the finite set of natural numbers , i.e. the natural numbers strictly below ; the left-hand sum runs over this set, and the right-hand double sum runs over the same set independently in each of its two variables, giving terms. The argument of the character on the right is the real difference , subtraction in (so it may be negative; this is not truncated natural-number subtraction), and is then applied to that real difference.
Degenerate case: when the index set is empty, both sides are the empty sum , the conjugate of is , and the statement reads . When it reads . The equality is an equality of complex numbers, asserted unconditionally for all and all simultaneously.
Two further definitions from this bundle are in scope but do not appear anywhere in the statement: the generating function of a finite set , given by , and the distance-to-the-nearest-integer function . Neither the sums here nor the character arguments are tied to , to any set , to , or to any rational approximation; the statement involves only the arbitrary sequence , the bound , and the character .
Confirmed by the mission captain (proposal self-audit).